Autophagy Regulates the Effects of Adipose-Derived Stem Cell Small Extracellular Vesicles On Lipopolysaccharide-Induced Pulmonary Microvascular Barrier Damage

Background: Small extracellular vesicles (sEVs) have been recognized to be more effective than direct stem cell differentiation into functional target cells in preventing tissue injury and promoting tissue repair. Our previous study demonstrated the protective effect of adipose-derived stem cells (ADSCs) on lipopolysaccharide (LPS)-induced acute lung injury and the effect of autophagy on ADSC functions, but the role of ADSC-derived sEVs (ADSC-sEVs) and autophagy-mediated regulation of ADSC-sEVs in LPS-induced pulmonary microvascular barrier damage remains unclear. Methods: After treatment with sEVs from ADSCs with or without autophagy inhibition, LPS-induced human pulmonary microvascular endothelial cell (HPMVECs) barrier damage was detected. LPS-induced acute lung injury in mice was assessed in vivo after intravenous administration of sEVs from ADSCs with or without autophagy inhibition. The effects of autophagy on the bioactive miRNA components of ADSC-sEVs were assessed after prior inhibition of cell autophagy. Results: We found that ADSC-sEV effectively alleviated LPS-induced apoptosis, tight junction damage and high permeability of PMVECs. Moreover, in vivo administration of ADSC-sEV markedly inhibited LPS-triggered lung injury. However, autophagy inhibition, markedly weakened the therapeutic effect of ADSC-sEVs on LPS-induced PMVECs barrier damage and acute lung injury. In addition, autophagy inhibition, prohibited the expression of ve specic miRNAs in ADSC-sEVs -under LPS-induced inammatory conditions. Conclusions: Our results indicate that ADSC-sEVs protect against LPS-induced pulmonary microvascular barrier damage and acute lung injury. Autophagy is a positive mediator of sEVs function, at least in part through controlling the expression of bioactive miRNAs in sEVs. Differences between groups were determined by one-way analysis of variance and post hoc Bonferroni corrections for multiple comparisons. The histologic semiquantitative analysis was compared using a nonparametric Mann–Whitney test. A P-value < 0.05 was considered to be statistically signicant. lung injury. Autophagy inhibition affected the expression levels of let-7-a-1, miR-21a, miR-143, miR-145a and miR-451a and mediated the protective effects of ADSC-sEVs against LPS-induced damage to the lung microvascular endothelial barrier. These results provide new insights into the roles and the related mechanisms of ADSC-sEVs in LPS-induced acute lung injury and suggest that autophagy regulation might be a potential strategy for modulating the treatment ecacy of ADSC-sEVs in lung injury.


Introduction
Sepsis-induced acute lung injury (ALI) is a major cause of acute respiratory distress syndrome, which is a major contributor to high morbidity and mortality. Pulmonary microvascular leakage is one of the characteristics of blood-air barrier dysfunction in ALI (1). In recent years, multiple experimental and clinical studies have been conducted to clarify the pathogenesis of ALI, and advances have been made in ALI treatment. However, few effective therapies have been developed to improve the outcome of ALI.
Stem cell-related treatments have been shown to be effective in treating injury and the repair of some organs. Adipose-derived stem cells (ADSCs) are a type of mesenchymal stem cell that have been identi ed as ideal candidates for cell-based therapies based on their relative abundance and easy accessibility (2). In addition, recent studies have shown that ADSCs have much stronger paracrine potential and tolerance under certain stress conditions than other types of stem cells (3,4). Paracrine components, especially exosomes, a type of small extracellular vesicles (sEVs), have been shown to be vital contributors to the e cacy of stem cell paracrine signaling. Exosomes, which are small membraned vesicles (30-100 nm), originate from multivesicular bodies formed by inward budding of the endosomal membrane. Exosomes carry complex biologically active components, including proteins, DNA, mRNA and lipids, among which miRNAs have been suggested to have an effective role in mediating exosome functions (5)(6)(7). However, there is still no method to isolate very pure exosomes, so the term "small extracellular vesicles (sEVs)" but not "exosomes" has been used in present study. Our previous study showed that ADSCs protect against lipopolysaccharide (LPS)-induced pulmonary microvascular barrier damage (8). However, the effect of ADSC-derived sEVs (ADSC-sEVs) under this condition is still unknown.
Autophagy is a protein and organelle degradation pathway that is pivotal for maintaining cellular homeostasis and promoting survival in response to stress conditions. Recently, the relevance of autophagy to sEVs has been tested. Autophagy affects the production of sEVs, which may be attributed to the link between small extracellular vesicle biogenesis and autophagy via the endolysosomal pathway, and these two processes share common proteins (9)(10)(11). In addition, autophagy is a major cellular degradation process that induces the degradation of various biological molecules, including proteins, lipids, and RNA, some of which are the bioactive components of sEVs (12,13). In a previous study, we found that autophagy regulated the release of certain growth factors from ADSCs in LPS-induced lung injury (8). Based on the aforementioned ndings, we hypothesized that autophagy regulates sEV function by regulating the bioactive components in sEVs. The goals of this study were to examine the function of ADSC-sEVs in LPS-induced pulmonary microvascular endothelial barrier injury and determine the role of autophagy in mediating the effects of ADSC-sEVs. were purchased from Sigma-Aldrich (Burlington, MA, USA). Endothelial cell growth medium (1001) was purchased from ScienCell (Carlsbad, CA, USA). A small-interfering RNA construct targeting autophagyrelated gene 5 (siATG5, sc-41446), an siRNA transfection reagent system (sc-41447), and an antibody against ATG5 (sc-133158) were purchased from Santa Cruz Biotechnology (Dallas, TX, USA). Antimicrotubule-associated protein 1-light chain 3 (LC3) B (3868), anti-Beclin-1 (3495), anti-GAPDH (2118) and anti-rabbit IgG (7074) antibodies were purchased from Cell Signaling Technology (Beverly, MA, USA).
Adipose-derived stem cell culture and treatment Human ADSCs, purchased from Cyagen Biosciences (Santa Clara, CA, USA), were cultured in DMEM. The primary cells were harvested when they had grown to approximately 80% con uence, and then the cells were plated on new culture dishes at approximately 6000 cells/cm 2 . To determine whether autophagy in uenced ADSC-sEV effects on LPS-induced microvascular barrier damage, we constructed ADSC lines with or without autophagy inhibition using an siRNA targeting ATG5. For siRNA transfection, 2 × 10 6 cells were transfected with 50 nM siATG5 using an siRNA transfection reagent system. After 36 h, the autophagy level of the cells was measured. Then, the cells were treated with interleukin (IL)-1β for 6 h, and sEVs were collected according to the undermentioned experimental method. ADSC siRNA-NC -sEVs represent sEVs derived from ADSCs transfected with negative control siRNA. ADSC siATG5 -sEVs and ADSC-sEVs represent sEVs derived from ADSCs with and without autophagy inhibition, respectively.

Isolation of sEVs
For small extracellular vesicle isolation, small extracellular vesicle concentration solution was used. Brie y, ADSCs were washed with PBS several times and cultured in DMEM supplemented with 10% exosome-free fetal bovine serum. After reaching con uence, the cells were treated with DMEM containing 1 ng/ml recombinant human IL-1β and incubated for 24 h. The culture medium was collected and centrifuged at 300 × g for 15 min at 4°C, followed by centrifugation at 2500 × g for 30 min. The supernatant was then ltered and mixed with small extracellular vesicle concentration solution at a 4:1 ratio, and the mixture was ultracentrifuged at 100,000 × g for 4 h at 4°C. Then, the pellets were overlaid on a 30% sucrose/deuterium oxide cushion and ultracentrifuged at 100, 000 × g for 1 h at 4°C. Finally, the extracted sEVs were collected and resuspended in 200 μl of PBS.
Human pulmonary microvascular endothelial cell culture and in vitro cell groupings Human pulmonary microvascular endothelial cells (PMVECs) (PromoCell, Heidelberg, Germany) were cultured in an endothelial cell medium. The cells were detached and transferred to new dishes at a split ratio of 1:2 for further propagation until they grew to con uence (usually 3-5 d). PMVECs at passages 3 to 5 were selected for analysis. PMVECs were divided into four groups as follows: PMVECs, LPSchallenged PMVECs, and LPS-challenged PMVECs cultured with ADSC-sEVs or ADSC siATG5 -sEVs. To mimic LPS-induced lung microvascular injury, PMVECs were incubated in endothelial cell medium supplemented with 10% fetal bovine serum containing 100 ng/ml LPS, followed by the addition of 20 μg/ml sEVs in 100 μl of PBS. The doses of sEVs were determined according to previous studies with minor modi cations (14,15). After 24 h, the cells were collected for further study.
Identi cation of ADSC-derived sEVs According to previous reports (16,17), transmission electron microscopy was used to observe the doublelayer ultrastructure of puri ed ADSC-sEVs. Nanoparticle tracking analysis was used to determine the average diameter and concentration of sEVs. The common sEV-speci c protein markers TSG101, CD9 and the nonspeci c protein CD68 were measured via western blotting.

Protein preparation and immunoblotting
ADSCs or PMVECs were homogenized in RIPA lysis buffer, and then the homogenate was incubated on ice for 45 min and centrifuged at 4°C (12,000 g for 5 min). After determining the protein concentration, the protein was collected and separated via sodium dodecyl sulfate polyacrylamide gel electrophoresis at 120 V for 2 h. The proteins in the gels were transferred onto a polyvinylidene di uoride membrane, which was then incubated with speci c primary antibodies, followed by incubation with horseradish peroxidaseconjugated secondary antibody for 1 h. Finally, protein visualization was performed using Pierce ECL western blotting substrate and autoradiography. The following primary antibodies were used: anti-LC3B, anti-Beclin-1, anti-ATG5, anti-ZO-1, anti-claudin-5, anti-TSG101, anti-CD9, anti-CD68, anti-Bcl-2, anti-Bax and anti-GAPDH. Quantity One 4.6 software was used to analyze the blots. The data were normalized to GAPDH and are expressed as the optical density (OD) integration.

Trans-endothelial permeability assay
PMVECs were cultured on the upper wells in a Transwell system, and FITC-dextran (1 mg/ml, MW 40,000) was added to the top of the wells and allowed to permeate through the PMVEC monolayer. After LPS treatment and ADSC-sEV culture for 6 h, the medium was collected from the lower compartments of the Transwell chambers and replaced with an equal volume of basal cell medium. The uorescence value of FITC-dextran in the medium was determined with a uorescence microplate reader (FLX800TBID, BioTek Instruments, Inc., Winooski, VT, USA) at an excitation wavelength of 492 nm and an emission wavelength of 520 nm.

Detection of PMVEC viability
We used an MTT assay to assess the viability of PMVECs. Each group was analyzed in triplicate at a density of 2000 cells/well. The cells were incubated with 5 mg/ml MTT during the last 4 h of LPS challenge. After removal of the supernatant, 100 ml of dimethyl sulfoxide was added to each well, followed by 10 min of shaking to dissolve the crystals. The OD of each well was measured at 490 nm with a spectrophotometer. The experiment was repeated three times in each group.
A LIVE/DEAD viability/cytotoxicity kit was used to further measure cell viability. Brie y, the cells were cultured on sterile glass coverslips as con uent monolayers. Then, 20 ml of 2 mM ethidium homodimer (EthD)-1 was added to 10 ml of PBS and combined with 5 ml of a 4 mM calcein AM solution. The working solution, which contained 2 mM calcein AM and 4 mM EthD-1, was directly added to the cells. After 15 min, the cells were examined using a confocal laser-scanning microscope.
Detection of apoptosis via ow cytometry An Annexin V-FITC apoptosis detection kit and ow cytometry were used to determine the apoptosis rate according to the manufacturer's instructions. Brie y, PMVECs were digested with 0.25% trypsin and then rinsed twice with PBS. Then, the cells were resuspended in 1× binding buffer at a concentration of 1×10 6 cells/ml, and 100 µl of the resuspended cell solution was transferred to 5-ml culture tubes. Then, 5 µl of Annexin V-FITC and 5 µl of propidine iodide were added to the culture tubes. The resulting solution was incubated at room temperature in the dark for 15 min, after which 400 µl of 1× binding buffer was added. The apoptosis rates were analyzed immediately via ow cytometry (BD Biosciences, San Jose, CA, USA).

F-actin labeling
We determined stress ber formation by measuring F-actin using a rhodamine-conjugated phalloidin molecular probe according to the manufacturer's instructions. Cells were treated with 100 ng/ml LPS and ADSC-sEVs, xed with 3.7% paraformaldehyde for 10 min, permeabilized with 0.5% Triton X-100, and nally stained with rhodamine-conjugated phalloidin. The nuclei were labeled with 4',6-diamidino-2phenylindole. The labeled cells were analyzed under a Nikon A1 R laser confocal microscope. We quanti ed F-actin levels in different groups by analyzing the percentage of cells containing stress bers.

Establishment of LPS-induced acute lung injury mouse models and study grouping
Eight-to ten-week-old wild-type BALB/c mice (Animal Center, Wenzhou medical University, Wenzhou, China) were used. The mice were starved of solid food but had free access to water 12 h before the experiments. All experimental protocols were approved by the Animal Care Ethics Committee of Wenzhou Medical University. All mice were handled in compliance with the Guidelines for the Care and Use of Laboratory Animals (18). The mice were anesthetized with pentobarbital sodium (50 mg/kg intraperitoneally), orally intubated with a sterile plastic catheter and challenged by intratracheal instillation of 2mg of LPS kg −1 b.w. ADSC-sEVs or ADSC siATG5 -sEVs (100 μg/ml, 200 μl total volume) were intravenously administered via a caudal venous canula 30 min following LPS challenge as described previously (15). The mice were sacri ced 48 h after LPS instillation.

Lung histopathology and edema detection
The mice were sacri ced 48 h after LPS instillation, and fresh left lung tissue was dissected and xed immediately in 10% formalin and then para n-embedded and cut into 4-μm thick para n sections. The sections were stained with hematoxylin and eosin (H&E). Lung injury was assessed according to four categories: interstitial in ammation, neutrophil in ltration, congestion, and edema. The assessment results are shown as scores on a 0-to 4-point scale: no injury = score of 0; injury in 25% of the eld = score of 1; injury in 50% of the eld = score of 2; injury in 75% of the eld = score of 3; and injury throughout the eld = score of 4 (19). Ten microscopic elds from each slide were analyzed. The sums of the tissue slides were averaged to evaluate the severity of lung injury. All microscopic sections were scored by a pathologist who was blinded to the experimental groups and protocol.
In addition, the right lungs of the mice were immediately removed, and the wet weight was measured after the experiment. Then, the same lungs were dried at 56°C for 72 h. The wet/dry lung weight ratio was then calculated to assess the severity of lung edema.

Protein concentration in bronchoalveolar lavage uid (BALF)
After the experiment, the left general bronchus of each mouse was ligated, and then the right lung was irrigated with 0.5 ml PBS after LPS challenge. The uid from three lavages was pooled. The total BALF from each mouse was centrifuged at 4°C (1000 rpm for 15 min). The protein concentration of the BALF supernatants was determined using a mouse total protein S Kit according to the manufacturer's instructions.

Detection of cytokines via ELISA
The expression of the in ammatory cytokines TNF-α and IL-1β in the BALF was measured using ELISA kits in accordance with the manufacturer's instructions.
Quanti cation of ve speci c miRNAs in sEVs using real-time RT-PCR Total RNA was isolated from sEVs using a GenElute Mammalian Total RNA Kit according to the manufacturer's instructions. cDNA synthesis was executed using a PrimeScript reverse transcription reagent kit with gDNA eraser. Reverse transcription was performed using a One Step PrimeScript™ RT-PCR Kit. The sequences of the forward primers used are shown in Table 1.

Results
Effective inhibition of autophagy using siATG5 ATG5 is indispensable in both canonical and noncanonical autophagy. Through siATG5 treatment, we effectively reduced autophagy levels. Western blotting demonstrated that the expression of ATG5 was most effectively diminished in siATG5 439-transfected ADSCs ( Figure 1A), and thus, siATG5 439 was selected to inhibit autophagy in subsequent experiments. In addition to the expression of ATG5, that of LC3-II and Beclin-1, two other essential autophagy proteins, was markedly inhibited by siATG5 ( Figure 1B, C). Morphological assessment via transmission electron microscopy showed that autophagosomes were double-or multimembrane structures that engulfed cytoplasmic components ( Figure 1D). Statistically, the number of autophagosomes per mm 2 of cell cross section in the siATG5-treated group was signi cantly lower than that in the control group ( Figure 1E).

Isolation and characterization of ADSC-derived sEVs
Previous studies have shown that preconditioning mesenchymal stem cells with cytokines or speci c conditioned medium can enhance their paracrine functions, including the effects of exosomes on tissue injury and repair (20)(21)(22). In this study, we preconditioned ADSCs for 6 h with IL-1β, one of the vital proin ammatory cytokines induced by LPS, and then collected ADSC-derived extracellular vesicles using a small extracellular vesicles concentration solution kit and supercentrifugation. Western blotting demonstrated the presence of the small extracellular vesicle marker proteins TSG101 and CD9 but the absence of CD68 in these vesicles (Figure 2A). In addition, the isolated ADSC-derived extracellular vesicles ranged in size from 70-120 nm, as determined by nanoparticle tracking analysis, and IL-1β preconditioning promoted the production of these extracellular vesicles ( Figure 2B). Transmission electron microscopy analysis showed that isolated ADSC-derived extracellular vesicles had a typical cupshaped morphology in both the control and IL-1β preconditioning groups ( Figure 2C). These ndings indicated that these vesicles ful lled the minimal experimental criteria for sEVs (17). Therefore, these vesicles are referred to as ADSC-small extracellular vesicles (sEVs). We collected ADSC-sEVs from IL-1βpreconditioned ADSCs for further experiments.

Autophagy inhibition reduced the protective effect of ADSC-sEVs on the expression of tight junctionrelated proteins
To further test the effect of autophagy on sEV function in LPS-induced pulmonary microvascular barrier damage, we extracted equal concentrations of sEVs from ADSCs in the presence or absence of autophagy inhibition. Then, these sEVs were added to PMVECs in the presence of LPS. We found that LPS inhibited the expression of ZO-1 and claudin-5, two critical tight junction-related proteins in PMVECs. However, ADSC-sEV treatment, signi cantly inhibited this change in PMVECs, and autophagy inhibition weakened the effect of ADSC-sEVs on the expression of ZO-1 and claudin-5 ( Figure 3A, B).
Autophagy inhibition reduced the protective effect of ADSC-derived sEVs on PMVEC apoptosis and viability PMVEC apoptosis has been used as one of the critical assessment indices for LPS-induced pulmonary microvascular barrier damage (23). Flow cytometry showed that LPS markedly increased the percentage of endothelial cell apoptosis, which was effectively reduced by ADSC-sEVs. However, autophagy inhibition, signi cantly weakened the function of ADSC-sEVs ( Figure 4A, B). In addition, we measured the expression of Bax and Bcl-2, which are classic pro-and antiapoptotic proteins, respectively. LPS promoted the expression of Bax and reduced the expression of Bcl-2; ADSC-sEVs inhibited the expression of Bax and promoted that of Bcl-2 under LPS stimulation. Autophagy inhibition weakened these effects of ADSC-sEVs (Figure 4 C, D).
Cell viability was measured to further test the effect of autophagy on sEV function. A LIVE/DEAD viability/cytotoxicity kit was used to investigate cell viability. As shown, LPS treatment signi cantly increased the percentage of dead cells, which were characterized by PI staining of the nuclei. Small extracellular vesicle pretreatment markedly alleviated LPS-induced cell death. However, autophagy inhibition markedly weakened sEV-mediated abrogation of cell death (Figure 5 A, B).
In addition, an MTT assay was used to further test cell viability. LPS signi cantly reduced cell viability, which was apparently alleviated by ADSC-sEVs. Autophagy inhibition reduced the effect of sEVs ( Figure  5C).
Autophagy inhibition reduced the protective effect of ADSC-sEVs on pulmonary microvascular permeability Microvascular permeability has been used as one of the representative indices to assess pulmonary microvascular barrier integrity (24). In this study, we found that LPS stimulation for 6 h or 12 h increased microvascular endothelial cell permeability, which was signi cantly reduced by sEV treatment. Autophagy inhibition markedly weakened the effect of ADSC-sEVs on LPS-induced microvascular permeability ( Figure 6).

PMVECs
Previous studies have shown that LPS induces F-actin polymerization to form contractile actin bundles and stress bers. Contraction of stress bers leads to the formation of intercellular gaps that increase the permeability of the endothelial barrier (25,26). To test whether LPS-induced stress ber formation could be regulated by sEVs, we incubated endothelial cells with sEVs from ADSCs with or without autophagy inhibition under LPS stimulation. As shown, LPS signi cantly increased the formation of actin stress bers, and this effect was signi cantly inhibited by sEVs. However, autophagy inhibition reduced the effect of sEVs on stress ber formation ( Figure 7A). We further quanti ed the percentage of cells containing stress bers in the different groups. LPS treatment markedly increased the proportion of cells containing stress bers, which was effectively decreased by ADSC-sEV treatment. However, autophagy inhibition weakened the effect of ADSC-sEVs on the formation of stress bers ( Figure 7B).

Regulation of autophagy affected the treatment e ciency of ADSC-sEVs in LPS-induced acute lung injury
LPS stimulated a striking in ux of polymorphonuclear leukocytes into the alveolar space, as well as marked congestion and edema in the lung tissue. Administration of ADSC-sEVs effectively attenuated LPS-triggered lung injury. The effect of ADSC siRNA-NC -sEVs on lung injury was similar with that of ADSC-sEVs. However, the protective effect of ADSC siATG5 -sEVs was much worse than that of ADSC-sEVs or ADSC siRNA-NC -sEVs ( Figure 8A). The severity of lung injury was also assessed using a semiquantitative histopathology scoring system, ADSC-sEV treatment effectively decreased LPS-induced lung injury. However, inhibition of autophagy, markedly weakened the therapeutic e cacy of ADSC-sEVs ( Figure 8B).
In addition to morphologic evidence, lung edema was assessed by detecting the lung wet/dry weight ratio. LPS challenge increased the lung wet/dry weight ratio, which was effectively attenuated by ADSC-sEV treatment. However, the inhibitory action of ADSC siATG5 -sEVs on LPS-induced lung edema was much weaker than that of ADSC-sEVs ( Figure 8C).
In vivo lung microvascular permeability was assessed by detecting the total protein concentration in BALF. As shown, LPS markedly increased protein leakage in BALF, which was signi cantly attenuated by ADSC-sEV or ADSC siATG5-NC -sEV treatment. However, the inhibitory effect of ADSC siATG5 -sEVs on protein leakage in BALF was markedly weaker than that of ADSC-sEVs ( Figure 8D).
Regulation of autophagy affected the treatment e ciency of ADSC-sEVs in the LPS-induced excessive in ammatory reaction.
An exaggerated in ammatory response is one of the causes of LPS-associated ALI. To evaluate the in ammatory reaction, we measured the levels of TNF-α and IL-1β, two important proin ammatory cytokines, in BALF. As shown, LPS signi cantly increased the production of TNF-α and IL-1β, which was effectively inhibited by ADSC-sEV or ADSCs siATG5 -sEV treatment. However, the inhibitory effects of ADSC siATG5 -sEVs on the production of TNF-α and IL-1β were signi cantly weaker than those of ADSC-sEVs ( Figure 9A, B).

Autophagy affected the expression of speci c miRNAs in ADSC-sEVs
To test the effect of autophagy on bioactive components transferred by ADSC-sEVs, we detected the expression changes of speci c miRNAs (let-7-a-1, miR-21a, miR-143, miR-145a and miR-451a) that have been found in ADSC-sEVs (27). We measured the expression pro le of the aforementioned miRNAs in ADSC-sEVs with or without autophagy inhibition. IL-1β treatment increased the expression of miR-21a and decreased that of let-7-a-1, miR-143 and miR-145a but did not affect the expression of miR-451a. Interestingly, autophagy inhibition weakened the expression of all these miRNAs under IL-1β stimulation ( Figure 10).

Discussion
In this study, we found that ADSC-sEVs protect against LPS-induced pulmonary microvascular barrier damage and acute lung injury by alleviating apoptosis and reducing the loss of the tight junction-related proteins ZO-1 and claudin-5. Autophagy is one of the essential regulators of the protective effect of ADSC-sEVs by affecting the expression pro les of at least the aforementioned ve speci c miRNAs within ADSC-sEVs.
sEVs are one of the pivotal components of stem cell paracrine signaling and have been shown to be much more effective in some organ injuries and repairs than direct stem cell differentiation. Under normal conditions, most cells can secrete sEVs; however, pathogens or other stress stimuli may promote sEV secretion and/or alter sEVs contents (28)(29)(30). Hypoxic preconditioning enhanced the protective effect of bone marrow stromal cell-derived exosomes against acute myocardial infarction (31). Ischemic preconditioning can potentiate the protective effect of marrow stromal cell-derived exosomes on endotoxin-induced acute lung injury (32). In addition, LPS pretreatment not only induced exosome secretion by macrophages but also enhanced the effects of macrophage-derived exosomes on the proliferation and activation of hepatic stellate cells (33). Similarly, in the present study, we showed that preconditioning with IL-1β, one of the key proin ammatory factors induced by LPS, promoted the production of ADSC-sEVs and affected the expression of miRNAs in sEVs. Based on these ndings, IL-1β preconditioning is a viable option to enhance sEV functions and prevent LPS-induced lung injury.
In the present study, we found that sEV treatment signi cantly reduced endothelial cell apoptosis, which is one of the classic characteristics of LPS-induced endothelial barrier damage. Our ndings are consistent with those of previous studies. The administration of exosomes to staurosporine-treated Chinese hamster ovary cells effectively alleviated apoptosis and enhanced cellular viability (34). In a skin lesion model, ADSC-Exos inhibited HaCaT cell apoptosis and promoted cell proliferation to accelerate cutaneous wound healing (35). However, sEVs released from different cell types have different biological effects, and different stress stimuli may trigger different functions in homologous sEVs. Some researchers have found that tumor-derived exosomes carrying immunosuppressive factors can induce apoptosis in activated CD8+ cells and NK cells to suppress immunotherapy e cacy (36). These ndings suggest that the effects of sEVs on target cell apoptosis are not uniform and that the sEV origin and pathological conditions may be key regulatory factors.
The bioactivity of sEVs is ultimately attributed to their protein and nucleic acid components. miRNAs are the most numerous cargo molecules in sEVs; they are selectively sorted into sEVs and transferred to recipient cells, where they mediate certain target mRNAs and cell functions. In the present study, we found that stimulation with IL-1β increased the expression of miR-21a and decreased that of let-7-a-1, miR-143 and miR-145a, but did not affect the expression of miR-451a. These ndings suggest that more than one miRNA participates in regulating the effects of sEVs in alleviating LPS-induced endothelial barrier damage. Although many previous studies have highlighted the pivotal effects of sEV miRNAs, many of these studies focused on the function of speci c miRNAs. Our ndings suggest that sEV functions are likely to be due to the cooperative effects of various miRNAs. It is essential for us to implement further studies to clarify the relevance and crosstalk among at least the aforementioned four speci c miRNAs that are altered in ADSC-sEVs under IL-1β conditions. Autophagy, which is a lysosomal-dependent degradation and recycling pathway, has traditionally been suggested to maintain protein, lipid and organelle homeostasis. Recently, autophagy has been identi ed as one of the vital mediators of sEV biogenesis and function. We found that the same concentration of sEVs collected from ADSCs in the presence or absence of autophagy inhibition had different protective effects on LPS-induced pulmonary microvascular endothelial barrier damage and lung injury. Autophagy inhibition partially weakened the protective effect of ADSC-sEVs, as indicated by increases in the apoptosis rate and stress ber formation but reduced expression of tight junction-related proteins in endothelial cells. These data provide extremely strong evidence suggesting that autophagy can affect sEV functions. Our ndings are consistent with those of previous studies. Autophagy regulation modulates the effect of retinal astrocyte-derived exosomes on the proliferation and migration of endothelial cells (37). On the one hand, autophagy shares molecular machinery with exosome biogenesis, and there is substantial crosstalk between these two processes (38). In addition to its traditional roles in maintaining protein, lipid and organelle homeostasis, increasing evidence indicates that autophagy can impact RNA homeostasis. Beyond its other degradative capabilities, autophagy can degrade RNA, RNAbinding proteins and ribonucleoprotein complexes (39,40). In the present study, we found that autophagy inhibition lowered the expression of let-7-a-1, miR-21a, miR-143, miR-145a and miR-451a under IL-1β stimulation in ADSC-sEVs. This result likely explains why autophagy affects sEV functions. Further studies are essential to classify the mechanism by which autophagy mediates sEV miRNA expression.

Conclusion
In conclusion, we showed that ADSC-sEVs were bene cial in maintaining pulmonary microvascular barrier integrity and lung injury. Autophagy inhibition affected the expression levels of let-7-a-1, miR-21a, miR-143, miR-145a and miR-451a and mediated the protective effects of ADSC-sEVs against LPSinduced damage to the lung microvascular endothelial barrier. These results provide new insights into the roles and the related mechanisms of ADSC-sEVs in LPS-induced acute lung injury and suggest that autophagy regulation might be a potential strategy for modulating the treatment e cacy of ADSC-sEVs in lung injury.       The permeability of PMVECs was measured using a Transwell assay. LPS increased the permeability of endothelial cells, and this effect was signi cantly alleviated by ADSC-sEVs treatment. Autophagy inhibition weakened the protective effect of ADSC-sEVs against LPS-induced endothelial permeability.
The experiment was repeated three times.

Figure 7
The effect of ADSC-sEVs with or without autophagy inhibition on stress ber formation.  The effect of ADSC-sEVs with or without autophagy inhibition on LPS-induced acute lung injury.
(A) Lung histology visualized with hematoxylin and eosin staining. (B) Microscopic injury of the lung was statistically scored. The results are presented as the mean ± SD. (C) Lung water content was tested by detecting the wet/dry weight ratio. (D) Lung microvascular permeability was assessed by detecting the changes in total protein concentration in bronchoalveolar lavage uid (BALF). The results are presented as the mean ± SD (n = 18/group, 6 for H&E staining and pathological scores; 6 for wet/dry weight ratio; 6 for BALF collection and assessment of total protein level).

Figure 9
The effect of ADSC-sEVs with or without autophagy inhibition on LPS-induced in ammatory reactions in mouse lungs.
(A-B) ELISAs were used to detect the concentrations of TNF-α and IL-1β in BALF from mouse lungs that underwent LPS treatment. The results are presented as the mean ± SD (n = 6/group) Figure 10 Autophagy mediates the miRNA expression pro le of ADSC-sEVs. The expression of let-7-a-1, miR-21a, miR-143, miR-145a and miR-451a in ADSC-sEVs. The results are expressed as the mean ± SD of three independent experiments.